A lithium iron phosphate battery cell processing data management method and system

Abstract

The application discloses a lithium iron phosphate battery cell processing data management method and system, and belongs to the technical field of battery manufacturing and data management. The method comprises the following steps: collecting, in real time, battery cell processing data covering the whole process of electrode sheet manufacturing, battery cell assembling and formation and capacity grading at a frequency of not less than 1 time per second; assigning a unique identity code to each battery cell and performing data correlation; storing the encrypted data by using a layered architecture, and optionally storing the data shards in a blockchain network; performing real-time quality monitoring and early warning on the whole process data based on preset rules; and displaying the whole process data graph in response to a traceability request. The application also provides a corresponding system, electronic equipment and storage medium. Through whole-process high-frequency data binding, intelligent preprocessing, real-time monitoring and secure storage, the application realizes single battery cell level accurate traceability, early abnormality warning and cross-subject data auditing, and significantly improves the process control level and data governance capability of lithium iron phosphate battery cell manufacturing.

Description

Technical Field

[0001] This invention relates to the fields of battery manufacturing technology and data management technology, and more specifically, to a method, system, electronic device, and computer-readable storage medium for managing data throughout the entire process of lithium iron phosphate battery cell manufacturing. Background Technology

[0002] Lithium iron phosphate (LFP) cells have been widely used in new energy vehicles, energy storage systems, and other fields due to their advantages such as high safety, long cycle life, and low cost. However, the manufacturing process of LFP cells is extremely complex, encompassing multiple steps including electrode fabrication (mixing, coating, rolling), cell assembly (winding / stacking, electrolyte injection), and formation and capacity testing. The process parameters in each step (such as slurry particle size, coating surface density, compaction density, electrolyte injection volume, formation voltage and current, etc.) have a decisive impact on the final cell performance (such as capacity, cycle life, and safety).

[0003] Traditional data management methods for battery cell production suffer from the following drawbacks: First, data from each process is stored in isolation on different devices or systems, forming "data silos" that make it difficult to achieve data traceability throughout the entire lifecycle of a single battery cell. Second, the low frequency of data collection, poor standardization, and simplistic handling of missing and outlier values ​​lead to unreliable data analysis results. Third, the lack of a real-time quality monitoring mechanism means that abnormal parameters are often only discovered after finished product testing, resulting in significant waste. Fourth, data storage security is insufficient, especially when multiple responsible parties (such as foundries, equipment manufacturers, and OEMs) are involved, making data susceptible to tampering and difficult to audit. Therefore, there is an urgent need for a highly reliable, traceable, and intelligently monitored data management method that covers the entire LFP battery cell process. Summary of the Invention

[0004] The present invention aims to solve the problems existing in the prior art and provide a method and system for managing lithium iron phosphate battery cell processing data.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for managing lithium iron phosphate battery cell processing data includes the following steps:

[0007] Data Acquisition Steps: During the processing of lithium iron phosphate (LFP) battery cells, sensors, industrial cameras, and PLC controllers deployed on coating machines, winding machines, liquid injection machines, formation cabinets, and dimensional inspection stations are used to collect cell processing data in real time at a frequency of no less than once per second. The cell processing data covers the entire process from electrode fabrication and cell assembly to formation and capacity testing, and includes at least the following: slurry particle size and solid content data from the mixing process; areal density and dimensional data from the coating process; thickness and compaction density data from the rolling process; alignment and tension data from the winding / stacking process; liquid injection volume and vacuum level data from the liquid injection process; and voltage, current, and capacity data from the formation and capacity testing processes.

[0008] Data preprocessing and association steps: The collected raw data is cleaned, deduplicated, completed, and standardized to establish a standardized dataset. A unique QR code or RFID identification code is assigned to each lithium iron phosphate cell, and the identification code is bound and associated with the full-process processing data corresponding to that cell.

[0009] Data storage steps: The associated cell processing data is encrypted and stored using a hierarchical architecture. Structured process parameters and test results are stored in a relational database, while equipment logs, test videos, and raw curve data are stored in a time-series database or a non-relational database.

[0010] Quality monitoring steps: Based on preset process parameter ranges and process logic rules, the entire process data of a single battery cell is compared and analyzed in real time to identify out-of-limit data and data inconsistencies that violate process logic. When an anomaly is identified, a quality warning message is generated, containing the battery cell's unique identification code, the abnormal process, and a description of the anomaly.

[0011] Data service steps: In response to a traceability request, retrieve and display the corresponding full-process processing data map from the database based on the target identification code.

[0012] As a further improvement of the present invention, in the data preprocessing and association steps, the 3σ criterion combined with the process experience threshold is used to identify and remove abnormal data, and missing data is filled by interpolation based on historical data sequences. The standardization process includes data normalization and unit of measurement unification. The unique identification is a QR code or radio frequency identification (RFID) tag.

[0013] As a further improvement of the present invention, the unique QR code or radio frequency identification (RFID) identification code is set on the battery cell casing or tab by laser engraving or affixing, and is identified by automatic reading equipment in subsequent processes to achieve real-time binding of data with the physical battery cell.

[0014] As a further improvement of the present invention, the data storage step further includes: dividing the encrypted associated data into multiple data fragments according to the production process nodes; and distributing the multiple data fragments among multiple nodes of a blockchain network jointly maintained by the equipment suppliers, manufacturers, and quality inspectors of each process, so as to ensure the immutability and auditability of the data throughout the entire process during the cross-responsibility transfer process.

[0015] As a further improvement of the present invention, the method further includes a performance prediction step: training a cell performance prediction model based on historical processing data and corresponding finished cell performance test results, wherein the performance test results include at least cycle life, rate performance and safety test results; using the trained cell performance prediction model, performing preliminary performance screening on cells produced online, and marking cells with predicted performance below a preset threshold as key testing targets.

[0016] The present invention also provides a lithium iron phosphate battery cell processing data management system for implementing the method described in any of the above-mentioned embodiments, the system comprising:

[0017] Data acquisition module: Used to receive raw processing data from sensors, industrial cameras and PLC controllers deployed in each process at a preset frequency.

[0018] Data processing and association module: Used to clean and standardize raw data, and generate and manage unique identification codes for battery cells, thereby binding data with identification codes.

[0019] Data storage module: Includes relational databases, time-series databases, or non-relational databases, used to store associated data in a hierarchical architecture.

[0020] Quality monitoring module: Used to load process rules, perform compliance and logical analysis on real-time or historical data of the entire battery cell process, and generate quality warning information.

[0021] Data Service Module: Used to respond to internal and external requests, providing data traceability, visualization, and report generation services based on identity codes.

[0022] As a further improvement to the system, the data storage step may also include fragmenting the encrypted associated data and storing it on a blockchain network to ensure its immutability and auditability.

[0023] As a further improvement to the system, the system also includes: a performance prediction module, used to store historical data and performance mapping relationships, train and apply cell performance prediction models, and output a list of objects to be tested; an identification management and reading module, used to control the generation and attachment of identification codes, as well as the automatic reading devices configured at each node of the production line; and an access control unit, used to assign different data access, operation, and analysis permissions according to user roles.

[0024] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0025] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] 1. Full-process traceability: By assigning a unique identification code to each cell and binding it with high-frequency data from electrode manufacturing to capacity testing, a full life-cycle data map at the single-cell level is constructed, which greatly improves the accuracy and efficiency of locating quality problems.

[0028] 2. High data quality: The data was cleaned using a high-frequency acquisition strategy (≥1 time / second) combined with the 3σ criterion and process experience thresholds, as well as an intelligent interpolation completion method, which ensured the accuracy, completeness and consistency of the dataset, laying a solid foundation for subsequent analysis.

[0029] 3. Real-time monitoring and early warning: Based on preset rules, the system performs online compliance and logical analysis on the entire process data, which can generate early warning information in the early stages of anomalies, facilitating timely intervention and reducing scrap rate.

[0030] 4. Secure and reliable storage: It adopts a layered architecture (relational + time-series / non-relational database) to balance query efficiency and unstructured data storage needs; it introduces a blockchain sharding storage mechanism to ensure data immutability and full auditability in scenarios involving multiple responsible parties.

[0031] 5. Intelligent Expansion: It integrates performance prediction models and production statistical analysis functions, enabling online preliminary screening of battery cell performance and quantitative evaluation of production process capabilities, providing data-driven decision support for process optimization and intelligent manufacturing. Attached Figure Description

[0032] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0033] Figure 1 This is a flowchart of a lithium iron phosphate battery cell processing data management method according to an embodiment of the present invention.

[0034] Figure 2 This is a module structure diagram of a lithium iron phosphate battery cell processing data management system according to an embodiment of the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example 1

[0037] This invention provides a method for managing data during the processing of lithium iron phosphate battery cells.

[0038] 1. Data Collection

[0039] On a lithium iron phosphate battery cell production line, an online particle size analyzer and viscometer are installed at the dual planetary mixer (mixing process); a β-ray areal density meter and a CCD dimensional detection camera are installed behind the transfer coating machine; a thickness laser sensor is installed at the roller press; a tension sensor and alignment detection system are installed at the winding machine; a mass flow meter and vacuum gauge are installed at the liquid injection machine; and voltage and current data for each channel are directly read inside the formation and capacity testing cabinets. All sensor and camera data are aggregated by a PLC and uploaded to a central data acquisition server at a frequency of 1Hz (1 time / second). The acquired data includes: slurry D50 particle size, solid content, coating areal density (positive / negative electrode), electrode width, roller press thickness, winding alignment, electrode tension, liquid injection volume, vacuum degree, formation voltage curve, and capacity testing, etc.

[0040] 2. Data Preprocessing and Association

[0041] A unique QR code (Data Matrix code) is laser-engraved on the tabs before the battery cell is assembled. An industrial barcode scanner is installed at the entrance of each subsequent process.

[0042] After receiving the raw data, the acquisition server performs cleaning and completion.

[0043] Outlier elimination — combination of 3σ criterion and process experience threshold: taking the rolling thickness data as an example. Let the continuous collected thickness data sequence of a certain cell in the rolling process be x1, x2, …, xn. First, calculate the arithmetic mean μ and standard deviation σ of this sequence:

[0044]

[0045] According to the 3σ criterion, if a certain data point xi satisfies , it is regarded as an outlier caused by random error and eliminated. However, in the production of LFP cells, the actual allowable fluctuation range of some parameters may be much narrower than the statistical distribution range (for example, the process specification requires the thickness to be 150±3μm, while the statistical 3σ range may reach 150±5μm). Therefore, the present invention further combines the process experience threshold: if xi exceeds the absolute upper and lower limits set by the process (for example, <147μm or >153μm), even if it is still within the 3σ range, it is forcibly marked as abnormal and eliminated.

[0046] Missing data imputation — interpolation method based on historical data sequence: For missing values caused by short-term scanning code failure or instantaneous sensor disconnection, linear interpolation or spline interpolation is adopted

[0047] to perform imputation. Let the valid data points adjacent before and after the missing value be (t0,y0) and (t1,y1), then the imputed value y at the missing time t (where t0<t<t1) is:

[0048]

[0049] For data with periodic characteristics (such as the periodic fluctuation of coating areal density in the web width direction), the weighted moving average method of historical concurrent data is adopted for imputation.

[0050] After that, all data are unified into the International System of Units, and min-max normalization processing is performed. The processed data is associated with all process data of the cell through the two-dimensional code information read by the scanner.

[0051] 3. Data storage

[0052] The associated structured data (such as process parameter values, pass / fail marks) is encrypted and stored in the MySQL relational database. Unstructured or semi-structured data such as the original equipment logs of each process, the voltage-time curve (CSV format) of the formation process, and the vacuum degree change curve of the liquid injection process are encrypted and stored in the InfluxDB time series database.

[0053] For critical data involving joint production by the contract manufacturer and the brand owner (such as formation capacity data), the following further steps are taken: the data is divided into 7 data segments according to the process of "mixing-coating-rolling-winding-liquid injection-formation-capacity testing," and stored in 4 nodes of the Hyperledger Fabric blockchain network jointly maintained by the contract manufacturer, the brand owner, the cell material supplier, and the third-party testing agency. The hash value of each data segment is recorded on the blockchain, and the original data is held by the authorized party, ensuring that no party can unilaterally tamper with historical data.

[0054] 4. Quality Control

[0055] Deploy a rules engine in the central data processing server. Preset rules include, for example, setting the upper limit for the areal density of the coating process to 160 g / m². 2 The lower limit is 140 g / m 2 The liquid injection volume deviation must not exceed ±1% of the target value; process logic rules include: "The liquid injection valve must not be opened before the vacuum degree of the liquid injection process reaches -90kPa." When the system detects in real-time comparison that the liquid injection volume of a certain cell is 135g (3% lower than the target value), an early warning is immediately generated: "Cell ID: LFP202410001, Abnormal Process: Liquid Injection, Abnormal Description: Liquid injection volume is lower than the lower limit of -3%, please check the liquid injection pump." This warning is sent to the engineer on duty via MES system pop-up and SMS.

[0056] 5. Data Services

[0057] When an end user or quality inspector scans the QR code on the battery cell casing, the system receives a traceability request. Based on the QR code ID, the data service module retrieves all parameters, curves, and keyframes of the inspection video from the relational database and time-series database for that battery cell, from mixing to capacity testing, and displays them on the web interface as a timeline graph. Clicking on any anomaly point directly links to the original equipment log.

[0058] Example 2: Detection of winding anomalies and precise cell-level tracing based on "alignment and tension data"

[0059] This embodiment demonstrates in detail how to use the method of the present invention to discover and locate a hidden anomaly in actual production.

[0060] Background: In the winding process, the overhang between the positive and negative electrodes and the separator, as well as the electrode tension, are key parameters that determine the risk of internal short circuits and interface performance of the battery cell. Excessive overhang can lead to lithium plating, while excessive tension fluctuations can cause electrode wrinkling.

[0061] Anomaly detection process:

[0062] During a night shift production period, the system collected the winding data of a batch of battery cells (batch number: BATCH-241101) in real time. The data processing and association module generated a unique QR code for each battery cell (such as LFP241101-0001 to LFP241101-5000).

[0063] The process rules loaded by the quality monitoring module are as follows:

[0064] Rule 1 (Over-limit rule): The allowable alignment range is [-0.5mm, +0.5mm], and an alarm will be triggered if it is exceeded.

[0065] Rule 2 (Logical Rule): Tension Volatility = (Maximum Tension - Minimum Tension) / Average Tension. If the volatility is >15% and the duration exceeds 2 seconds, it is considered abnormal.

[0066] When the system reached the 1247th cell (ID: LFP241101-1247), it analyzed the alignment data stream in real time and found that the negative extreme alignment drifted from the normal +0.2mm to -0.7mm within 3 seconds, triggering rule 1. At the same time, the tension sensor data of this cell showed that the tension fluctuation rate reached 22% and lasted for 2.5 seconds, triggering rule 2.

[0067] The quality monitoring module immediately generated an early warning message: {“Cell ID”: “LFP241101-1247”, “Abnormal Process”: “Winding”, “Abnormal Code”: “OVR-ALIGN-01”, “Abnormal Description”: “Negative extreme alignment exceeds the lower limit of -0.7mm, and tension fluctuation rate exceeds the standard”, “Timestamp”: “2024-11-01 02:35:22”}. This warning was sent to the engineer on duty via MES system pop-up and SMS.

[0068] Traceability and Location: After receiving the alert, engineers traced the battery cell by inputting "LFP241101-1247" into the data service module. The system displays a complete data graph of the cell's lifecycle, highlighting key features:

[0069] 1. Winding pattern: The alignment and tension curves show synchronous and violent fluctuations between 02:35:20 and 02:35:25 seconds.

[0070] 2. Equipment-related data: The system automatically retrieves the servo motor encoder feedback data and take-up shaft vibration spectrum data of the winding machine (equipment ID: W-03) within the same time period.

[0071] 3. Root cause analysis: Through comparison, it was found that the vibration frequency of the winding shaft bearing of the W-03 machine showed a characteristic spike at abnormal times, which was consistent with the tension fluctuation frequency.

[0072] Result: Engineers immediately stopped the machine and inspected the W-03 machine, discovering wear on the ball bearings of the take-up shaft. After replacing the bearings, the alignment and tension data of subsequent battery cells returned to normal. Utilizing the traceability function of this invention, the time from the occurrence of the anomaly to locating the specific physical spare part was only 15 minutes. In contrast, in the traditional model, it is necessary to wait until the finished battery cells undergo short-circuit testing or aging to detect batch defects before disassembly and analysis, which typically takes several days to several weeks.

[0073] Example 3: Construction and Application of Cell Performance Prediction Model

[0074] This embodiment details the specific implementation of the performance prediction model described in claim 5.

[0075] 1. Model type selection:

[0076] This embodiment uses the Gradient Boosting Tree (XGBoost) model. XGBoost has advantages such as strong robustness, good interpretability, and fast training speed when processing high-dimensional, mixed-type (continuous values ​​+ categorical values) industrial data.

[0077] 2. Input feature vector construction:

[0078] Based on historical data and process experience, the following 20-dimensional feature vectors are extracted from the entire process data as model input:

[0079] Process Feature number Feature Name Explanation (Data Preprocessing Method) Mixing F1 Slurry D50 average particle size Intra-batch mean Mixing F2 CV value of slurry solids content Coefficient of variation reflects batch stability Coating F3 Average surface density of positive electrode Coating F4 Positive electrode surface density CPk Process capability index Coating F5 Negative electrode / positive electrode areal density ratio Key design parameters Roller pressing F6 compaction density Roller pressing F7 Extreme difference in electrode thickness Reflects the uniformity of roller pressing winding F8 Maximum alignment deviation absolute value winding F9 95th percentile of tension volatility winding F10 Electrode wrinkle index Extraction based on machine vision Injection F11 Relative error of injection volume (Actual - Target) / Target Injection F12 vacuum decrease rate Vacuum difference / time before and after injection form F13 First effect Initial discharge capacity / charge capacity form F14 lithium plating characteristic peak area Extract from dV / dQ curves form F15 Internal resistance growth rate slope of internal resistance change during formation process Capacity F16 room temperature discharge capacity Capacity F17 High-temperature aging capacity decay rate Full process F18 First pass rate of key processes Was the battery cell defective during a certain process? Full process F19 Device ID One-Hot Encoding Category characteristics, marking production machines Full process F20 Production period Early / Midday / Night Shifts, Category Characteristics

[0080] 3. Output labels (target variable):

[0081] Continuous value prediction: Predict the capacity retention rate (%) of the cell after 800 cycles under 1C / 1C conditions.

[0082] Classification and prediction: The battery cells are divided into three grades: "Excellent (≥90%)", "Good (80%-90%)", and "Poor (<80%)".

[0083] 4. Model training process:

[0084] Dataset: Historical full-process data and corresponding measured cycle life data of 50,000 battery cells produced in the past 6 months were collected.

[0085] Data partitioning: The data was randomly partitioned into training set (40,000), validation set (5,000), and test set (5,000) in a ratio of 8:1:1.

[0086] Training parameters: learning rate (eta) = 0.05, maximum tree depth (max_depth) = 6, subsample ratio (subsample) = 0.8, number of iterations (n_estimators) = 500. Early stopping is used; training stops when the validation set loss function does not decrease for 20 consecutive iterations.

[0087] Feature Importance Analysis: After training, the importance of the output features is ranked. In this model, the top five features are, in order: lithium plating peak area (F14), maximum alignment deviation (F8), cathode areal density CPk (F4), first-time efficiency (F13), and relative error of liquid injection volume (F11).

[0088] 5. Online Application (Initial Performance Screening):

[0089] The trained XGBoost model is deployed to the online inference service. For newly rated cells, the system automatically extracts their F1-F20 feature vectors and inputs them into the model. The model outputs the predicted cycle life retention rate of the cell.

[0090] Decision rules: If the predicted retention rate is less than 80%, the cell is marked as a "key inspection target," and the system automatically isolates it for more stringent EIS testing and longer-term random inspections; if the predicted retention rate is ≥ 90%, the inspection frequency can be relaxed. This mechanism increases the early interception rate of defective products by approximately 35%, while reducing unnecessary destructive testing costs.

[0091] In other embodiments, the performance prediction model may also be a random forest, a support vector machine (SVM), or a deep neural network (DNN).

[0092] Example 4: Cross-entity data storage and auditing based on consortium blockchain

[0093] This embodiment details the specific architecture and process of the blockchain storage steps.

[0094] Application Scenario: A high-end energy storage LFP cell is designed by Company A (the brand owner), manufactured by Company B (the contract manufacturer), with key electrode sheets supplied by Company C (the material supplier), and ultimately procured by Company D (the energy storage system integrator). All four parties must share the entire cell manufacturing process data, but no party may unilaterally tamper with it.

[0095] Blockchain architecture:

[0096] Network type: Consortium blockchain (using the Hyperledger Fabric framework).

[0097] Node configuration:

[0098] Endorsement nodes (4): Deployed on the internal servers of the four parties, A, B, C, and D respectively.

[0099] Ordering nodes (3): provided by an independent third-party cloud service provider, using the Raft consensus mechanism (fault tolerance: can tolerate a maximum of 1 node failure).

[0100] Committing nodes: All endorsing nodes act as committing nodes simultaneously.

[0101] Channel: Create a dedicated channel "LFP-Channel" that only allows nodes from parties A, B, C, and D to join.

[0102] Data upload process (taking capacity data as an example):

[0103] 1. Data generation: Inside Company B's factory, the formation cabinet completes the charging of cell ID "LFP241101-1247" and generates capacity data "Capacity=105.2Ah".

[0104] 2. Encryption and Segmentation: The system automatically associates this data item with the entire process data of the battery cell, and then calls the encryption module (AES-256) for encryption. According to the predefined smart contract rules, the encrypted data is divided into 7 data segments according to "mixing-coating-rolling-winding-liquid injection-formation-capacity grading".

[0105] 3. Initiating a transaction proposal: Company B's client application simultaneously initiates a transaction proposal to four endorsing nodes, which includes: the hash value of the data fragment, the timestamp, the cell ID, and the operator's signature.

[0106] 4. Endorsement: Each endorsing node simulates the execution of the smart contract (verifying data format and checking permissions), signs the result, and returns an "endorsement response".

[0107] 5. Submission and Ordering: After the client collects at least 3 consistent endorsement responses (satisfying 2f+1, where f=1), it packages the transaction and sends it to the ordering node. The ordering node performs a global sort on the transactions and packages them into blocks.

[0108] 6. Distribution and Verification: The sorting node broadcasts the new block to all committing nodes. After verifying the validity of the block, the committing nodes write it to their local ledger.

[0109] 7. Completion: Block on-chain. The original data fragments (encrypted) are stored by Company B in its local time-series database. Only the hash value, metadata, and storage location pointer are stored on the blockchain.

[0110] Audit process:

[0111] When Company D needs to audit the true capacity of a batch of battery cells, it can: 1) obtain the hash value H of the corresponding battery cell capacity data from the blockchain; 2) request the original encrypted data fragment from Company B; 3) calculate the hash value H' of the original data; 4) compare H and H' to see if they match. If they match, it proves that the data has not been tampered with; if they do not match, it indicates that the data provided by Company B is not authentic. Since the on-chain hash value is witnessed by all four parties and is tamper-proof, this mechanism provides a powerful third-party audit capability.

[0112] Comparison of Results: Key Indicators Before and After Applying This Invention

[0113] To quantify the "unexpected technical effects" brought about by this invention, a comparative test was conducted over a period of 6 months at a lithium iron phosphate battery cell factory with an annual production capacity of 2GWh (the first 3 months were conducted using the traditional method, and the last 3 months were conducted after applying the method and system of this invention). Key performance indicators are compared below:

[0114] Indicator Item Traditional model (before application) The invention mode (after application) Improvement range illustrate Quality traceability response time Average 4.2 hours Average 8.5 seconds Reduced to the second level From relying on manual record flipping and report checking to obtaining the entire process map simply by scanning a code. Abnormal batch location time Average 2.5 days Average 22 minutes Shortened by 99.4% By combining real-time monitoring and data correlation, it can quickly locate specific machines / time periods. Overall product defect rate 2.8% 1.6% Reduced by 42.9% Thanks to real-time early warning and early intervention, batch defects have been reduced. Lithium plating defect rate after formation 1.2% 0.4% Reduced by 66.7% Thanks to the closed-loop real-time monitoring of winding alignment and tension Overall Equipment Effectiveness (OEE) 78.5% 86.2% Increased by 9.8% Reduce unplanned downtime caused by quality issues Customer complaint closed-loop time An average of 7 working days Average 1.5 working days Shortened by 78.6% We can provide clients with an immutable, end-to-end data map for rapid self-verification. Cross-entity data audit time Not applicable (unauditable) or several weeks <5 minutes A qualitative leap Blockchain-based evidence storage enables third-party audits to be completed instantly. Performance prediction accuracy not applicable 92.3% - Predicted recall rate for "poor" grade battery cells (cycle life <2000 cycles)

[0115] Conclusion: As shown in the table above, this invention not only achieves a leap in traceability efficiency from "hourly" to "secondly," but also directly leads to a significant reduction in defect rates and an improvement in equipment efficiency. In particular, it reduces the lithium plating defect rate by 66.7%, which is difficult to achieve with simple statistical process control (SPC). The fundamental reason for this is that this invention achieves multi-parameter (alignment + tension + current) correlation analysis and real-time early warning at the single-cell level. These data fully demonstrate that the technological advancements achieved by this invention are significant and "unexpected." The data sharding storage mechanism of this invention is applicable to any distributed ledger technology.

[0116] Example 5

[0117] This embodiment provides a lithium iron phosphate battery cell processing data management system for executing the above method.

[0118] Reference Figure 2 The system includes:

[0119] Data acquisition module: Equipped with an OPC UA client, it communicates with the field PLC and pulls data at a frequency of 1Hz.

[0120] Data processing and association module: Deployed in the edge computing gateway, it performs data cleaning, normalization, interpolation completion, and calls the identifier management API to generate QR codes.

[0121] Data storage module: It consists of a master server running MySQL and three slave servers forming an InfluxDB cluster.

[0122] Blockchain storage module: Integrates Fabric SDK, responsible for data sharding, signing and submitting to the consortium blockchain.

[0123] Quality monitoring module: Runs the Drools rule engine to consume battery cell data streams from the Kafka message queue in real time.

[0124] Data service module: Provides RESTful API and front-end visualization page.

[0125] Performance prediction module: Includes an offline trained XGBoost model file and online inference service.

[0126] Identification management and reading module: includes laser marking machine, fixed barcode reader (Cognex) and RFID reader / writer head.

[0127] Access Control Unit: Based on the OAuth2.0 protocol, it distinguishes between roles such as administrators, process engineers, quality inspectors, and external customers to control the scope of data access.

[0128] The above system enables high-frequency, high-quality, safe, and intelligent management of the entire process data of lithium iron phosphate battery cells.

[0129] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing data during the processing of lithium iron phosphate battery cells, characterized in that, include: Data acquisition steps: During the processing of lithium iron phosphate cells, real-time data acquisition is conducted covering the entire process of electrode fabrication, cell assembly, and formation and capacity testing. Data preprocessing and association steps: The collected raw data is cleaned and standardized to establish a standardized dataset; a unique identifier is assigned to each lithium iron phosphate cell, and the identifier is bound and associated with the full-process processing data corresponding to that cell; Data storage steps: The associated cell processing data is encrypted and stored using a hierarchical architecture. Structured process parameters and test results are stored in a relational database, while unstructured data is stored in a time-series database or a non-relational database. Quality monitoring steps: Based on the preset process parameter range and process logic rules, the entire process data of a single battery cell is compared and analyzed in real time to identify abnormal data. When an abnormality is identified, a quality warning message containing the unique identifier of the battery cell, the abnormal process, and the description of the abnormality is generated. Data service steps: In response to traceability requests, retrieve and display the corresponding full-process processing data map from the database based on the target identity.

2. The lithium iron phosphate battery cell processing data management method according to claim 1, characterized in that, The data acquisition steps specifically include: Sensors, industrial cameras, and PLC controllers deployed on coating machines, winding machines, liquid injection machines, formation cabinets, and dimensional inspection stations are used to collect cell processing data in real time at a frequency of no less than once per second. The cell processing data includes at least the slurry particle size and solid content data of the mixing process, the areal density and dimensional data of the coating process, the thickness and compaction density data of the rolling process, the alignment and tension data of the winding / stacking process, the liquid injection volume and vacuum degree data of the liquid injection process, and the voltage, current, and capacity data of the formation and capacity testing processes.

3. The lithium iron phosphate battery cell processing data management method according to claim 1, characterized in that, The data preprocessing and association steps specifically include: Abnormal data is identified and removed using the 3σ criterion combined with process experience thresholds. Missing data is filled using interpolation based on historical data sequences. The standardization process includes data normalization and unification of units of measurement. The unique identifier is a QR code or a radio frequency identification (RFID) tag.

4. The lithium iron phosphate battery cell processing data management method according to claim 1, characterized in that, The unique identification mark is set on the cell casing or tab by laser engraving or affixing, and is identified by automatic reading equipment in subsequent processes to achieve real-time binding of data with the physical cell.

5. The lithium iron phosphate battery cell processing data management method according to claim 1, characterized in that, The data storage step further includes: The encrypted associated data is divided into multiple data fragments according to the production process nodes; these multiple data fragments are distributed and stored in multiple nodes of the blockchain network to ensure the immutability and auditability of the data throughout the entire process of data transfer across responsible entities.

6. The lithium iron phosphate battery cell processing data management method according to claim 1, characterized in that, The method also includes a performance prediction step: Based on historical processing data and corresponding battery cell performance test results, a battery cell performance prediction model is trained. The performance test results include at least cycle life, rate performance, and safety test results. Using the trained battery cell performance prediction model, the performance of battery cells produced online is initially screened, and battery cells with predicted performance below a preset threshold are marked as key targets for testing.

7. A lithium iron phosphate battery cell processing data management system, used to implement the method as described in any one of claims 1 to 6, characterized in that, include: Data acquisition module: used to receive raw processing data from sensors, industrial cameras and PLC controllers deployed in each process at a preset frequency; Data processing and association module: used to clean and standardize raw data, and generate and manage unique cell identifiers to achieve data-identity binding; Data storage module: including relational databases, time-series databases, or non-relational databases, used to store associated data in a hierarchical architecture; Quality monitoring module: Used to load process rules, perform compliance and logic analysis on real-time or historical data of the entire battery cell process, and generate quality early warning information; Data service module: Used to respond to internal and external requests, providing data traceability, visualization and report generation services based on identity.

8. The lithium iron phosphate battery cell processing data management system according to claim 7, characterized in that, Also includes: Blockchain storage module: used to shard the encrypted data and distribute it across blockchain network nodes; Performance prediction module: used to store historical data and performance mapping relationship, train and apply cell performance prediction model, and output a list of objects to be tested. Identification Management and Reading Module: Used to control the identification generation and attachment equipment, as well as the automatic reading equipment configured at each node of the production line; Access control unit: Used to assign different data access, operation and analysis permissions according to user roles.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6.

Images